synset

A synset, short for synonym set, is a fundamental unit in lexical databases such as WordNet. It represents a single lexical concept by grouping together words (lemmas) that can substitute for each other in many contexts without substantially changing the meaning. Each synset corresponds to one sense of a word, and is described by a brief definition called a gloss, often accompanied by example sentences.

Synsets are organized by part of speech and linked to other synsets through semantic relationships. Notable relations include hypernymy (is-a), hyponymy (type-of), meronymy (part-of), holonymy (whole-of), and antonymy. These links form a network that captures hierarchical and relational aspects of meaning, enabling comparisons of related concepts and the propagation of information through the structure.

A word may have multiple senses, each represented by a distinct synset. For example, the noun sense of “car” meaning a road vehicle can be captured by a synset that includes lemmas such as car, auto, and automobile, with a gloss like “a motor vehicle designed for transporting people on roads.” Other senses of the same word would be represented by additional, separate synsets.

Synsets provide a stable, language-wide representation of meaning that supports natural language processing tasks such as word sense disambiguation, information retrieval, machine translation, and measures of semantic similarity. They concentrate meaning in concept-centered units rather than surface word forms.